Walker, S.I. & Davies, P.
Journal of the Royal Society Interface, 10(79), 20120869.
Listen to the author interview! [mp3 download]
The origin of life has been a difficult subject to grapple. The primary challenge is that without a proper definition for life – a notoriously challenging problem in its own right – the problem of how life emerges is not well defined. Here we propose that the emergence of life may correspond to a transition in information processing capabilities from a purely trivial and passive, “bottom-up” flow of information to including “top-down” non-trivial flow that then actively influences the lower-level dynamics.
Life is a notoriously difficult concept to define. In most areas of biology, this is not a cause for concern – we can easily study our subject matter without recourse to defining it. However, in addressing life’s origin, applying at least a semi-rigorous definition of life becomes a critical issue. Of primary concern is that we do not know where to draw the line between very complex chemistry and very simple biology.
Darwinian evolution, or the evolution of genetic traits by natural selection, is agreed upon as being a fundamental characteristic of living systems. Because of the immense challenge presented by addressing the origin of life from a scientific viewpoint, it is typical to assume that if a simple chemical system capable of Darwinian evolution can be built, the rest is easy and the question of the origin of life will be solved. Thus, although few are likely willing to accept a simple replicating system of molecules (such as a growing crystal) as living, most researchers accept that Darwinian evolution will invariably lead to something everyone would agree is “alive”. This framework has led to traditional “genetics-first” models for the origin of life, such as the RNA world, that rely on the early appearance of a primitive genetic molecule capable of variation, selection, and heritage. “Metabolism-first” scenarios provide a contrasting viewpoint, which focuses on understanding how a select set of self-sustaining chemical reactions might emerge out of a prebiotic soup. The origin of life community has therefore tended to split into two camps, loosely labeled as “genetics-first” and “metabolism-first”, which, at least at first glance, appear to be at odds as to which came first.
The duality presented by the genetics/metabolism-first debate raises the question of whether either, by itself, is sufficient to provide a satisfactory account of the origin of life. A deep conceptual challenge arises in unifying the two approaches because they both are descriptions of chemical “hardware”, and neglect to aptly address the unique mode of information processing characteristic of living systems – i.e. the “software”. A distinguishing feature of biological information is that information is delocalized (the software is not necessarily with the hardware) and contextual (the software responds to the natural environment). The canonical example is provided by the information stored in DNA. DNA acts a passive data bank in the cell, storing information which is only active when read-out and implemented in the broader context of the cell. However, a cell has more information than stored in DNA alone. For example, an organism must inherit fully functional translational machinery for the information inherited in the genome to mean anything at all. The “algorithm” for cellular operation is distributed throughout the biochemical networks within the cell – including the current state of protein expression, non-coding RNAs and conditions supporting the self-assembly of protein and lipid structures. That’s not the whole story because in biology, information appears to have causal efficacy. This means that information stored in the current state of the system (DNA + biochemical network) not only determines the dynamics of the cell, but is also affected by the natural environment. Thus biological information is most aptly described as being active.
This active aspect of biological information leads to a very different causal narrative than that presented in traditional approaches to dynamical systems, where the dynamical laws are fixed and only the state evolves in time (Newtonian mechanics, for example, maps the state of the solar system today onto its state tomorrow using a fixed rule, and the new positions of the planets do not affect the rule). In this case, information plays a passive, unidirectional role. In biology, both the genetics-first and metabolism-first approaches to the origin of life typically fall within the realm of passive information processing and don’t account for how systems capable of active information processing emerge from systems that process information only passively. This is for good reason; the problem of how contextual information was able to be processed, and how it emerged in the first place from “mere bits” is formidable. Nevertheless, we propose that it is this transition that is the key defining characteristic of what is implicitly referred to when one speaks of the “emergence of life”: the emergence of context-dependent causal influences (i.e. the “software” can be modified by the natural environment – its context). In principle, such a transition should be well defined, marked by the transition from bottom-up to top-down causation, which should be mediated by a reversal in the dominant direction of the flow of information from lower to higher levels of organization (bottom-up), to that from higher to lower levels of organization (top-down). In such a transition, context therefore emerges as a new causal agent in a self-organizing chemical system.
We suggest that this transition may therefore be characterized by a transition in causal structure, where determining which phase (nonlife or life) a given system is in requires dynamical information and therefore can only be inferred by identifying causal relationships. This means that one cannot observe a potentially living system at one instant and decide whether it is living or not. We discuss one potential measure of such a transition, which is amenable to laboratory study and may provide a useful measure of systems approaching the living state by assessing how information is (or is not) dictating the dynamics of a given chemical system. In short, it may provide a means for distinguishing “just” complex chemistry from life. Much work remains to be done in this direction.
Find the paper in the ArXiv